Google CEO Admits AI Programming Lags Due to Lack of Developer Tools

ME AI News, according to monitoring by Beating, Google CEO Sundar Pichai, in an exclusive interview with The New York Times, acknowledged for the first time that Google is currently lagging behind industry leaders in advanced AI programming, tool usage, and long-range task execution. Pichai attributed this gap to Google’s lack of direct access to developers’ daily workflows through editors like Cursor or Claude Code, resulting in the loss of critical dynamic workflow data. However, his statement about “not yet opening a public interface” does not fully align with the actual market progress of Google’s AI programming product, Antigravity. In fact, Google launched its first-generation Antigravity IDE to the public market as early as November last year, but due to operational failures—including quota lock bugs, frequent server overloads and crashes, and restrictions on third-party open-source API access—the initial public beta failed to establish a stable developer trust ecosystem, causing many early users to migrate to competitors. During the interview, Pichai emphasized internal test data from Antigravity’s closed-loop operation at Google—such as weekly internal token consumption doubling and building an operating system from scratch in just 12 hours under extreme performance tests. By highlighting these internal metrics, Google aims to rebuild market confidence in its underlying agent programming capabilities, signaling that its core technological foundation remains robust, and its public-market setbacks stem primarily from limited scope in application scenarios and feedback loops. Last week’s urgent release of Antigravity 2.0 has now entered full public beta, attempting to win back developers’ attention through technical strength and incentive-driven moves. The high-profile promotion of internal achievements alongside the admission of missing public access is fundamentally intended to build industry confidence in Antigravity 2.0’s comeback, thereby establishing a real-world feedback loop connecting global developer input with algorithmic iteration. (Source: BlockBeats)